Shulou(Shulou.com)06/03 Report--

Http://blog.sina.com.cn/s/blog_cfa68e330102zg9l.html

Author | Vikram Mahidhar translator | Sambodhi Editor | Natalie

Introduction: Tao Zongyi in the late Yuan and early Ming Dynasty wrote in the South Village: there are birds in Wutai Mountain, which are called cold bugles. When in the heat of summer, his literary style was gorgeous, and he said to himself, "the Phoenix is not as good as me!" Compared to the severe cold in the depths of winter, the hairiness fell off, just like a brood, so he said to himself, "muddle along."

This is the prototype of the current second-grade text "the Story of the Cold Bird". In today's surging artificial intelligence, some companies' attitude towards artificial intelligence is not a "cold bird"? We are already living in the age of artificial intelligence. This is the best of times and the worst of times. The greatness of this era is that it is always updating and moving forward, but the sadness of this era is that those who fail to keep up with the times may never be able to keep up with them.

Although some companies, such as most large banks, Ford and GM, Pfizer (Pfizer, a multinational pharmaceutical company based in New York), and almost all technology companies, are actively embracing artificial intelligence. However, there are many companies that do not do so. Instead, they are waiting for the day when the technology is mature and artificial intelligence expertise is more widely used. They are all planning to be a "rising star", which is a strategy to cooperate with most information technology.

We don't think this is a good idea. It is true that some technologies do need further development, but some technologies (such as traditional machine learning) are quite mature and have existed in some form for decades. More recent technologies, such as deep learning, are based on research in the 1980s. New research has been carried out, but the mathematical and statistical basis of artificial intelligence has been established.

System development takes time

In addition to the issue of technology maturity, there are several other issues: the idea that companies can quickly adopt once the technology becomes more capable. First of all, it takes time to develop artificial intelligence systems. If these systems are fully generic, they may add little value to your business, so it takes time to customize and configure them according to your business and its specific areas of knowledge. If your artificial intelligence uses machine learning, then you have to collect a lot of training data. If it manipulates languages, like natural language processing applications, it may be more difficult to get the system up and running. Taxonomy and local knowledge need to be integrated into artificial intelligence systems, similar to the old "knowledge engineering" activities used in expert systems. This type of artificial intelligence is not only a problem of software coding, but also a problem of knowledge coding. It takes time to discover, disambiguate, and deploy knowledge.

In particular, if the supplier or consultant does not model your area of knowledge, it will usually take months for the architect. This is especially true for complex areas of knowledge. For example, the Catelyn Cancer Research Center (Memorial Sloan Kettering Cancer Center), in collaboration with IBM, has been using Waston to treat some types of cancer for more than six years, and despite high-quality talent in cancer care and artificial intelligence, the system is still not ready for widespread use. Some areas and business issues require the necessary knowledge engineering. However, it still needs to be manipulated according to the specific business environment of the company.

Integration takes time.

Even if your system has been built, there is still a problem of integrating artificial intelligence systems into your organization. Unless you are using some artificial intelligence features embedded in existing applications already used by the company (such as Salesforce Einstein in CRM), it takes a lot of planning and adaptation time to adapt to your business processes and IT architecture. The transition from pilots and prototypes to artificial intelligence production systems can be difficult and time-consuming.

Even if your organization is good at migrating pilots and prototypes to production, you must redesign your business processes to have a comprehensive impact on your business and industry. In most cases, artificial intelligence supports a single task, not the entire business process. Therefore, you need to redesign the business process and the new human tasks for it. For example, if you want to influence customer participation, you need to develop or adjust multiple artificial intelligence applications and tasks related to different aspects of marketing, sales, and service relationships.

Human-computer interaction in the era of artificial Intelligence

Finally, artificial intelligence needs to overcome human challenges. Few artificial intelligence systems are fully autonomous, and they pay more attention to the enhancement of artificial intelligence and the work done by human beings. New artificial intelligence systems often mean that people who work with them need to change new roles and acquire new skills. Moreover, it usually takes a long time to retrain employees to familiarize themselves with new processes and systems. Investment consulting firms that provide "robotic advice" to clients, for example, often try to get human consultants to turn their attention to "behavioural finance" or to provide advice and "nudges" to encourage informed investment decisions and actions. But this technique, unlike offering advice on buying tickets and bonds, takes some time to inculcate.

Even if the goal of an artificial intelligence system is to achieve full autonomy, it may take some time to enhance the mode. During this period, the key part of machine learning is realized through the interaction between the system and human users and observers. This is interactive learning, which is a key step for organizations to understand how the system interacts with its ecosystem. They can usually collect new data sets and begin to convert them into algorithms during that time, which usually takes months or years.

Management time of artificial intelligence applications

Although the purpose of the application of artificial intelligence systems is to provide exponential scaling and prediction, they require a new management method, which is more extensive than the traditional control and test-driven methods. The efficiency of artificial intelligence algorithms decreases over time because they are based on historical data and recent business knowledge. When the machine learns from the patterns in the new data, the algorithms can be updated, but they need to be monitored by subject matter experts to ensure that the machine can correctly interpret changes in the business environment. The algorithm must also continuously monitor the deviation. For example, if the artificial intelligence system is trained to create product recommendations based on customer demographics, and the demographics in the new data have changed significantly, biased recommendations may be provided.

Management also includes monitoring customer fraud. As the system becomes intelligent, users will become smarter. They may try to play with these systems with fraudulent data and activities. Monitoring and preventing customer fraud requires the deployment of complex instruments and manual monitoring in your business environment.

Winner takes all.

Therefore, it may take a long time to develop and fully implement an artificial intelligence system, and there are few shortcuts to the necessary steps. Once successfully implemented, scale can be very rapid-especially if the company has a rich supply of data and knowledge engineering. When late adopters have just completed all the necessary preparations, early adopters have already occupied a sizeable market share-they will be able to operate at lower cost and better performance. In short, there may be a situation in which the winner will always win, and the latecomers may never catch up. For example, think of a company like Pfizer, whose learning experience and capabilities, according to the director of its analytical and artificial intelligence lab, has accumulated more than 150 artificial intelligence projects in progress. Technology companies like Alphabet have more learning experience, with 2700 ongoing artificial intelligence projects as early as 2015.

Admittedly, if the company is willing to sacrifice its unique knowledge and way of doing business, some steps can be accelerated by waiting. Suppliers are developing a variety of knowledge maps and models that use technologies ranging from natural language processing to computer vision. If you have such a problem in your industry or business and are willing to adopt it without making any changes, you can speed up the adoption of artificial intelligence. However, if you don't adapt to your environment and build everything around it, you may lose your unique ability or competitive advantage.

Obviously, if you want to succeed in artificial intelligence and think there may be threats from artificial intelligence-driven competitors or new entrants to the market. Then you should start learning how to adapt it to your business in a variety of different applications and artificial intelligence methods. Some leading companies have created centralized artificial intelligence teams to do this on a large scale. These core teams focus on problem formulation, proof of business assumptions, modularization of artificial intelligence assets for reusability, creation of techniques for managing data pipelines, and cross-business training. Another possibility is to buy a startup that has accumulated a lot of artificial intelligence capabilities, but you still need to apply them to your business. In short, if you haven't started using artificial intelligence technology yet, what you should do is: start now! I hope it's not too late.

Original text link:

Https://hbr.org/2018/12/why-companies-that-wait-to-adopt-ai-may-never-catch-up

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.